-4 gets my vote.
Need to repeat to see if it's repeatable.-
Having trouble using the site?
Contact support
Load Development - Process
- Thread starter rlownote
- Start date
Need to repeat to see if it's repeatable.
Unless your rifle was locked into a mechanical rest and you were using something mechanical and repeatable to pull the trigger the same exact way each and every time you've got absolutely zero data.
Congratulations on burning up a bunch of components learning that the human is unreliable as fuck.
I settled on -0.006" and it has proven to be VERY repeatable. Coyotes at 800 plus!! I can ring the 4" steel at my pond boringly at 650 yards. If i miss, it was me missing it. Don't get me wrong, I miss it.-4 gets my vote.
Interesting at -0.007" the cloverleaf pattern of 3 groups.
I rarely shoot this many rounds finding my load but this rifle has such low recoil and is so extremely accurate, I just had to delve deeper. Unfortunately, I do not have pics of the rest of the load development but I tried more powder and more seating depths. I think I fired about 300 rounds in total seeing if there might be something I might be missing.
I'm not one that chases the burn. Once I find a good load, I shoot it for the life of the barrel.
My 280 Ackley is another that I burned some components on, only to go back to what I found after about 50 rounds (after the max charge testing).
I have 2 300WINMAGS with identical chambers. I found the combo that I am still using in both after only 50 rounds. Seating depth was my only change, again, after finding max charge. I have tried increasing powder charge and shit a few times over the decades but I always go back to what was true at the beginning.
I want accuracy first and foremost. However, I prefer that accurate shot to be as fast as possible, hence my testing for max charge.
I believe the point he was trying to make is that there isn't enough data available to adequately discern what variables (inputs) led to what specific outcomes (outputs).
-6 looks real good to me too, in fact, I was was debating between -6 and -4, but I liked the vertical in -4 discounting the couple that hit at 10 o'clock, simply attributing them to shooter's error. When do that kind of interpretation, it's the following verification test(s) that gives me a better idea if I'm right.
Undoubtedly the driver in play. When is the driver not in play?-6 looks real good to me too, in fact, I was was debating between -6 and -4, but I liked the vertical in -4 discounting the couple that hit at 10 o'clock, simply attributing them to shooter's error. When do that kind of interpretation, it's the following verification test(s) that gives me a better idea if I'm right.
I mean, do you go lock your rifle in a rest and wait for the coyote to come in to the exact spot or what? HAHA!! Yeah, go fuck yourself (not you, but whoever) with the remote trigger bullshit.
Yes, the driver is ALWAYS in play. I can almost always call my shots, especially the missed ones...check what I posted about the -0.007" group. I have fucked up once or twice in my life. That's why I have had to spend hours and hours finding an elk, a few deer, and a bear. I knew when the shot rang out, I had pulled the shot (the elk moved but that is beside the point). Shit happens.
EVERYONE misses.
EVERYONE is less accurate than the machine.
But to say that those groups are not enough data to decide what is a better seating depth than another is retardation defined.
It would be interesting to see what the results would be if you loaded up ammo from the best and worst recipes, and shot 5x5's with them.
I'd be curious to see if the same conclusions would be drawn as from the initial test.
There are articles written about a guy that had access to a real long warehouse to shoot in. No wind. They would clamp the rifles in vices and shit and use remote triggers to shoot. They had some amazing groups. I wish I could find an article. @sirhrmechanic may be able to point us there. I applaud their efforts. They figured out things like gluing the action to the stock would help in repeatability.
I would welcome any such research on maintaining a specific charge and only changing the seating depth.
I would also welcome any such research on holding the seating depth constant while varying the charge weight.
Can anyone provide this?
What they didn't figure out is how much the driver has to do with the system.
I don't care if your 50 pound 22-50BMg can hold 0.025 MOA in a rest using an air trigger. That's cool and all, but it's just not what I would ever use so it's not what I would consider.
It's the entire system. Take one of those warehouse rifles and pass it around. 10 world class shooters won't get the same 10 shot groups. They obviously didn't shoot enough rounds to verify the data. Right?
EDIT: Found one..https://precisionrifleblog.com/2013/10/18/secrets-of-the-houston-warehouse-lessons-in-extreme-rifle-accuracy/
While that is definitely accurate shooting, it's not in my personal realm of reality.
I would welcome any such research on maintaining a specific charge and only changing the seating depth.
I would also welcome any such research on holding the seating depth constant while varying the charge weight.
Can anyone provide this?
What they didn't figure out is how much the driver has to do with the system.
I don't care if your 50 pound 22-50BMg can hold 0.025 MOA in a rest using an air trigger. That's cool and all, but it's just not what I would ever use so it's not what I would consider.
It's the entire system. Take one of those warehouse rifles and pass it around. 10 world class shooters won't get the same 10 shot groups. They obviously didn't shoot enough rounds to verify the data. Right?
EDIT: Found one..https://precisionrifleblog.com/2013/10/18/secrets-of-the-houston-warehouse-lessons-in-extreme-rifle-accuracy/
While that is definitely accurate shooting, it's not in my personal realm of reality.
Last edited:
Are you talking about the infamous Houston warehouse?
Article here: https://www.angelfire.com/ma3/max357/houston.html
Much about what we use to build the accurate rifle has been gleaned from such ventures. When I was a teen, free floating the barrel was where it was at. Then "glass bedding" was the new and greatest shit. Then came adding an aluminum pillar to the bedding. Now a chassis is the shit.
That's all about the delivery device and fine and dandy. But this thread was started about the unit being sent down range, unless I am mistaken.
Load development for the sending device.
Where are the articles that delve into seating depth vs charge weight and all that shit and verifiable evidence....minus the driver?
EDIT: I will maintain my educated guess that a certain combination of barrel and projectile will be optimized for repeatability at a certain pressure. That pressure can be found at a given seating depth and powder charge. The same pressure can be found using a different seating depth and powder charge (node?...wtf ever) with very similar repeatability. Find a pressure in between and you have erratic results, even if the pressure is held steady. Why? It's not in the cards for that barrel and projectile. Harmonics is what I would consider the culprit.
That's all about the delivery device and fine and dandy. But this thread was started about the unit being sent down range, unless I am mistaken.
Load development for the sending device.
Where are the articles that delve into seating depth vs charge weight and all that shit and verifiable evidence....minus the driver?
EDIT: I will maintain my educated guess that a certain combination of barrel and projectile will be optimized for repeatability at a certain pressure. That pressure can be found at a given seating depth and powder charge. The same pressure can be found using a different seating depth and powder charge (node?...wtf ever) with very similar repeatability. Find a pressure in between and you have erratic results, even if the pressure is held steady. Why? It's not in the cards for that barrel and projectile. Harmonics is what I would consider the culprit.
Last edited:
I totally agree, the driver is always in play. And I'll add that aside from the skill of the driver, the gun will perform very different when locked in some sort of a vise vs. how it performs in the hands of a driver. The gun itself will react very different depending on how it's held.
That's why a big part of the shooting skill is being able to hold the gun the same every time the trigger is pulled.It's late and I need to hit the hay.
There is good, reliable data showing that seating a bullet against the lands creates greater pressure than seating deeper.
If you find your max pressure against the lands with x grains of powder and then seat the bullet deeper, your max pressure will be found with x plus y grains of powder. Seat deeper, you need ever more powder to find that max pressure.
To a point.
If you seat a bullet something like a half inch into the case, you will achieve max pressure again with less powder than 3/8"seating depth. It's actually pretty eye opening.
I don't even care to go find that for you guys.
There is good, reliable data showing that seating a bullet against the lands creates greater pressure than seating deeper.
If you find your max pressure against the lands with x grains of powder and then seat the bullet deeper, your max pressure will be found with x plus y grains of powder. Seat deeper, you need ever more powder to find that max pressure.
To a point.
If you seat a bullet something like a half inch into the case, you will achieve max pressure again with less powder than 3/8"seating depth. It's actually pretty eye opening.
I don't even care to go find that for you guys.
Guarantee you suck at prs too even after spending all that time on skill instead of load development.
Let’s also not forget the big bold face lies you’ve been telling in here so quite frankly no one gives a shit about your opinion.
I just did something like that for fun,
I took 100 rds of factory berger 140 gr hyb 6.5cm ammo and pulled half the bullets and seated them at a depth a KNOW my rifle likes with SMKs. Shot 10 5 shot groups of each.......
Well @rlownote, I'm sure this has been quite the ride for you!
Good luck on your journey, I'm sure its as clear as ever for you on what you need to do
Good luck on your journey, I'm sure its as clear as ever for you on what you need to do
Seating Depth is this too much?
I’m new to reloading any advice would help. 6.5 Creedmoor 140g Nosler, using H4350 up to 40.8g. Am I just asking for pressure issues seating the bullet this deep?
www.snipershide.com
What seating depth is actually doing...actual node or just pass/fail?
Not sure where to begin...but jump testing is something we all see a lot, and it typically the discussion goes back to bullet timing/barrel vibrations/etc...along those lines. one thing that has always hung in my mind is "forgiving" bullets...how are bullets "forgiving" if the whole idea of...
www.snipershide.com
Last edited:
The answer is at least 30, more like 100. Shoot the 30 and see if you want to change anything and you can have decent confidence in what you're seeing, but if you really want to know if it's the gun and not you you need to shoot 100 (or more).
Google it: if you search "what is a statistically valid sample size" you'll get: Most statisticians agree that the minimum sample size to get any kind of meaningful result is 100. 30 is the bare minimum as: The rule of thumb is based on the idea that 30 data points should provide enough information to make a statistically sound conclusion about a population. This is known as the Law of Large Numbers, which states that the results become more accurate as the sample size increases.
I've told no lies and am mid-pack in PRS in the most competitive region while not having enough time to devote to being truly competitive due to the time commitment and having to juggle a few successful businesses that require me to be available 24/7.
The only thing that's shown to be even close to being guaranteed is that I'm getting under your skin. The truth hurts.
Frankly it sounds like you're busy and projecting your own time constraints on people with more free time. Good enough sounds good enough for you and probably where you should have started and finished in this thread.
It's not like I'm going against the grain and upsetting the apple cart just to make noise. But a lot of you Load development/OCW/node assholes are wasting a lot of guys' time and components trying to convince them to do the same fool's errands you do so you can feel better.
Make every round you load come out the same as the next, buy a good scale instead of a fancy press or dies, drop every charge to within a kernel of the next, and seat all your bullets so they're all within a few thou of each other, chase low SDs instead of bulllshit... and if you do all that and it still doesn't shoot, then look in the mirror, because it's you.
Make every round you load come out the same as the next, buy a good scale instead of a fancy press or dies, drop every charge to within a kernel of the next, and seat all your bullets so they're all within a few thou of each other, chase low SDs instead of bulllshit... and if you do all that and it still doesn't shoot, then look in the mirror, because it's you.
No. Wrong.
We want every round to do the same predictable thing. Doing that doesn't require stumbling on some magic recipe, it requires consistency and accepting that the human is the most unmanageable variable. That's it.
I'm not going to go line by line quoting you in this thread but you've posted some untrue things, whether you know it or not. I'll leave it at that.
If we are going to be pedantic about it, there is no definitional statistically valid sample size. It's just a question of how confident you want to be. 10 observations could be plenty if you're dealing with extremely stable variables. Think about it: how many times do you need to check a random bolt hole to be certain what thread pattern it is? Do you need to test 30 bolts? No. Because bolts are consistent enough and there's a big enough difference between bolt patterns that you only need to check once (maybe a couple to eliminate close metric/standard threads) to be sure.
Finding a significant difference between two loads can be found with 5 shots if the loads are both extremely consistent and the difference between them is large.
One of the reasons 30 has been a rule of thumb, I think, is because historically that's where t-tables stopped reporting values (probably to keep the tables readable on a single page). Past that point, it was generally just taken for granted that the difference between a t distribution and a normal distribution converge to a meaningless difference, so the bottom line on the table was for df=infinity. Technically, a t distribution only converges to a normal distribution in the limit (infinity). Thus, given you were working with one of those tables, if you make sure to have n>30, you just use the z row for everything. Ie, that was as precise as the table allowed.
This is why I mentioned that you need to start by asking what type of precision is required in the load and what your gun is probably capable of. If you have a good barrel (and you're a good shooter) and you just need 0.75-1MOA at 100yds, just pick good components and load them consistently. Load development is probably a waste of your time.
Notice my user name? HAHAHA Use Google...go to fucking school, take a few courses in math. Maybe statistics will be on your transcript, too.
There is no rule of thumb. There are mathematically sound methods used to determine minimum sample size.
How many rounds are you going to fire finding your low SD that is good enough, statistically, to be relevant? 30 because Google?
I found your test interesting and I have a few comments on the test and its results. The first is about your jam data. You are incrementing in 0.001" increments with is extremely small. I am assuming this is the "as loaded" calculated jams. My question would be is this the actual jamb that was shot. Did the bullet set back any in the chambering process?
The other concerns the groups and in particular the one from -.003 on. The 10 shots all seem to have the same or almost the same POI and the group sizes (with the exception of -0.005" and- 0.007").are about the same. What is going on with those groups? What its your take on what is happening?
Here is my take on the results. The only results we can really take are anecdotal since you have no real quantitative data. As for seating depth it appears that once you are initially at -0.003" you are likely jammed and that the the groups are similar from that point on with the exception of the -0.005" and -0.007" groups. Because those groups are shot in a separate "shooting session" and the groups on either side of it very similar I would say that this group is probably, but not definitively, the result of setup/shooter issues and not the load. (NOTE: You seem to be very good at consistency in your shooting). Regardless of the number of shots you are looking at only 2 random shots to define the results, with some guidance in the size of the obliterated aiming point. I think If this were my test I would shoot 5 shots at -0.005" and -0.007" and see if they repeat this behavior. I honestly think that -0.006" is a good place to load, depending on the results of the retest at -0.005 and -0.007.
How I would I have performed this test? Since the intent was to shoot 10 rounds for each group of jams I would have planned to determine the mean point of impact for each shot and and the mean radius of each group of 10 shots. To do this I would have shot each shot on a separate small target. A rectangle is a good aiming point (corner) and setting the POI away from the aiming point. This makes it easy to measure the vertical and horizontal offset of each shot. I can use the data to calculate the mean point of impact, mean radius, and standard deviation of the the radius. There are statistical tools (Welch's T Test for mean radius and F Test for standard deviation) to then allow comparison of the groups to determine if they are statistically different. It is in performing these analytical tests that statistical significance comes into play.
Sounds like a lot? Cal Zant's Precision Rifle Blog carried a great series on statistics a couple of years ago. Here is a link to the Executive Summary with links to the three articles.
Loading…
precisionrifleblog.com
The link below is to a site that has a lot of statistical test/calculators that are well documented and easy to use.
Statistics online - checks assumptions, interprets results
Statistical tests, charts, probabilities and clear results. Automatically checks assumptions, interprets results and outputs graphs, histograms and other charts.
www.statskingdom.com
Thanks for the input. Of course, these loads were developed about 4 years ago....so....I found your test interesting and I have a few comments on the test and its results. The first is about your jam data. You are incrementing in 0.001" increments with is extremely small. I am assuming this is the "as loaded" calculated jams. My question would be is this the actual jamb that was shot. Did the bullet set back any in the chambering process?
The other concerns the groups and in particular the one from -.003 on. The 10 shots all seem to have the same or almost the same POI and the group sizes (with the exception of -0.005" and- 0.007").are about the same. What is going on with those groups? What its your take on what is happening?
Here is my take on the results. The only results we can really take are anecdotal since you have no real quantitative data. As for seating depth it appears that once you are initially at -0.003" you are likely jammed and that the the groups are similar from that point on with the exception of the -0.005" and -0.007" groups. Because those groups are shot in a separate "shooting session" and the groups on either side of it very similar I would say that this group is probably, but not definitively, the result of setup/shooter issues and not the load. (NOTE: You seem to be very good at consistency in your shooting). Regardless of the number of shots you are looking at only 2 random shots to define the results, with some guidance in the size of the obliterated aiming point. I think If this were my test I would shoot 5 shots at -0.005" and -0.007" and see if they repeat this behavior. I honestly think that -0.006" is a good place to load, depending on the results of the retest at -0.005 and -0.007.
How I would I have performed this test? Since the intent was to shoot 10 rounds for each group of jams I would have planned to determine the mean point of impact for each shot and and the mean radius of each group of 10 shots. To do this I would have shot each shot on a separate small target. A rectangle is a good aiming point (corner) and setting the POI away from the aiming point. This makes it easy to measure the vertical and horizontal offset of each shot. I can use the data to calculate the mean point of impact, mean radius, and standard deviation of the the radius. There are statistical tools (Welch's T Test for mean radius and F Test for standard deviation) to then allow comparison of the groups to determine if they are statistically different. It is in performing these analytical tests that statistical significance comes into play.
Sounds like a lot? Cal Zant's Precision Rifle Blog carried a great series on statistics a couple of years ago. Here is a link to the Executive Summary with links to the three articles.
Loading…
precisionrifleblog.com
The link below is to a site that has a lot of statistical test/calculators that are well documented and easy to use.
Statistics online - checks assumptions, interprets results
Statistical tests, charts, probabilities and clear results. Automatically checks assumptions, interprets results and outputs graphs, histograms and other charts.
I am quite sure that when I wrote 0.000" the bullets were just BARELY touching and not jammed at all. I use a stripped bolt and dry erase marker to find that point.
As far as obliterating the aiming point....well, it may appear as that is the case but if you notice the targets are oriented in such a way as to be able to use your scope crosshairs to line the corners of the target up. Horizontal points of the target aligned with the horizontal crosshair while the vertical points of the target are aligned with the vertical crosshair. I'm not even looking at the "bullseye". Smart, huh? FAR more reliable than trying to aim at the center of a little box.
I have a BS Math and a MS in engineering. I am not real interested in reading more about statistics. In fact, other than reading what my chronograph spits out as SD, I don't give 2 shits about statistics while I am pulling the trigger. Gregor ate peas, I am a carnivore.
Honestly, any of these groups from -0.003" to -0.009" (measured as away from the lands, there were no jammed bullets) could probably have been used. I chose -0.006" just in case I wanted to chase the throat erosion, I might have some prolonged time with this barrel. I have never done that, chase that throat erosion, but why would I want to shoot myself in the foot?
Last edited:
One of the toughest and most humbling experiences in my life was being asked to stop what I was doing and teach.
It is an easy mistake to get going too fast and too deep into ballistics/math and forget the customer may not be a peer level scientist.
Sometimes your audience is Low Drag/High Speed, but sometimes they are recruits or kids with zero background.
It is easy to get into a mental state where the beginner's needs are such a distant memory that teaching rookies can be a humbling and challenging experience.
On this forum and in particular the reloading page, we have to remember to leave room for the zero level folks to ask questions without giving them (or each other) a bad time. YMMV Carry on.
It is an easy mistake to get going too fast and too deep into ballistics/math and forget the customer may not be a peer level scientist.
Sometimes your audience is Low Drag/High Speed, but sometimes they are recruits or kids with zero background.
It is easy to get into a mental state where the beginner's needs are such a distant memory that teaching rookies can be a humbling and challenging experience.
On this forum and in particular the reloading page, we have to remember to leave room for the zero level folks to ask questions without giving them (or each other) a bad time. YMMV Carry on.
Any.
#1 - I've not ever seen any evidence of .001" seating depth make any discernable difference.
#2 - one can't even have confidence they are in fact going .001" from the lands at a time, due to (measurement consistency abilities with bto, measurement inconsistency from the lands, and bullet consistency)
So , why even attempt to test it that way? A much more productive way would be to go at least .003" at a time, with 10 shot groups. Then you'd at least have some meaningful info.
BS Physics and MS Mechanical Engineering here. I kinda picked on your test because you put some good targets up and I took it as an opportunity make some very general observations and comments, not specifically to criticize what you had done. Things like the target POI is fine when you go about it the way you described and I have used that method myself. In your case I thought that was how you had shot the groups. As you noted I misread the jam/setback which I'll give myself a swift kick in the butt for.
As for statistics I spent part of my career as a test engineer and I always hated statistics and when using it was forced to follow Test Codes so most of the times it was cookbook. The shooting world has gone from one extreme to another with regards to testing and statistical significance. We used to work with three or five shots but now we say that that number is worthless and we should have 20, thirty, or more. The reality is three or five may be all that is needed based on what is being looked at and with the understanding of the limitations of small sample sizes and how larger sample sizes would affect the results of the small sample size tested.
Then why did you put your ridiculous test on here? You ignore what MUCH more experienced people are telling you and acting a fool. This isn't the pit. And you can't know that even was "zero" with your maker test.
Yes.
As was said about the bolt, or can be said about a sheet full of cookies. A sample of one may be plenty. I'd rather spend my time shooting than entering data into some statistics equation the results of which mean actually near nothing in the real world.
Can YOU make the shot with YOUR equipment? The answer to that question is the only one that holds any water.
What I find funny is how all you guys argue and say you would rather spend your time shooting than testing and loading.
I really just don't understand that. I would rather be reloading and then testing than just shooting. what is the point of shooting if it isn't to test something??? you guys just shoot for no reason other to hit a target?
I really just don't understand that. I would rather be reloading and then testing than just shooting. what is the point of shooting if it isn't to test something??? you guys just shoot for no reason other to hit a target?
Let me add here about the single sample point as being enough and relate it to more shooting lore...
When I zero a rifle for the first time, I shoot one round. That round is most likely to not impact as near the point of aim as I desire but I see no reason to increase the sample size. I adjust my scope. I look at the target using the ruler in front of my eye and move the turrets appropriately.
As amazing as it may seem, as outlandish as it sounds, I adjust my scope after a sample of one.
Interestingly, most often, the second shot impacts very close to the point of aim and is sometimes right on the money. Fire for effect.
Want to know another time I only use a sample of one? When I take a rifle out of the safe and go verify my zero. Far more often than not, the first is right on the money and I don't sample another round. Fire for effect.
I do have a couple of rifles that the first round, clean cold bore, is off the point of aim a bit. Even those are predictable and if the first round clean cold bore is right where the others have been over the years, I don't usually sample another round. Fire for effect.
This works out ok. I have been able to repeatedly take game with one round from a clean cold bore.
These arguments about load development go round and round in this forum. Half the time people agree with each other violently because they never defined what they mean when they say "load development". For me, I set the bullet 20 thou off and load up charge weights to see where pressure is and figure out what velocity I want to be at. Then I may explore different OAL's if I don't already have some good groups. But in like 20 thou increments. And that usually works pretty well with a custom cut rifled barrel, quality brass, and a Hodgdon powder like Varget, 4350, 4831SC, or H1000. I call that a form of load development but some people are referring to round robin tests and OCW rituals when they say "load development" and that's what their thinking of when they say it's voodoo. But when you're talking about loading outside the conditions of well built match rifles, Berger bullets, Alpha OCD brass, and H4350 I've absolutely seen combinations of cartridges, powder, bullets, charge weights, and OALs that don't shoot well. And that's when load development isn't a farce. Lately I've been shooting a 22GT for competition and I just can't go wrong with it. It doesn't seem like seating depth or charge weights matter. Berger 85.5's, H4350, Alpha brass, CCI450's in a 23lb rifle. It just shoots. Load to desired velocity and seat the bullet wherever you want because it shoots well anywhere. But I have a 7 WSM hunting rifle in a Broughton barrel using Winchester brass that I thought was a dud I tried 65gr of RL26. It will shoot 1/2 moa with that load but that rifle has shot some shit groups with other powders and bullets. I also trimmed a 9mm load from 2-3" at 25yds to consistently sub 1" with 4.8gr of Ramshot Silhouette and RMR 135gr MWs. Seating depths, charge weights, powder type, and bullets do matter. And you have to shoot them in a systematic series of combinations to identify which work well and which don't. That's load development to me. And Ive seen it matter. Are there some myths around certain load development techniques that are Rorsach tests? Yes, I think there are. But I don't think that means all forms of load development are myths. This is where I think trend-followers get caught up in trying to emulate their idols and parrot their talking points in interpreted extremes.
My biggest problem with testing seating depth in .001" increments (besides the sheer amount of barrel life and components used) is I don't want a load that is temperamental enough for a .001 change to make a significant difference. If .001 brings a shit load into a great load it's too damn picky for me. I want a load that is resilient enough for a 10 to 20 thou difference to not matter. But I do agree that trying to find, and load to, a .001 can be a bit of a snipe hunt, not to mention on how consistent the BBTO on a lot of bullets are at times.
It's a futile effort, but yes absolutely. IF, you could get components and actually load to .001" , using calipers no less..... then what would that even do for you? If there's a noticeable difference in .001" seating depth, why in the world would you want that to be your loaf? It just doesn't make sense, IF.... you could even do it reliably.
With todays equipment and components, I think many of us make good ammo not because of our process and "tests", but despite of it.
3-5 rounds for testing variables certainly "works" for what we are doing. Not because those are sample sizes statistically relevant to make the deductions we do, but rather because some of the variables we try to control and test just don't matter as much as we think they do.
1 5-round group isn't going to give enough information to conclude how one variable change affects the outcome, but a lot of times we prescribe too much influence on that variable we are trying to test anyways.
