Food tracking

laurachambers86

I used to use MFP all the time for tracking, recently decided to give a different app a go for a while as it also has coaching and I felt like I needed that boost. Anyway, I did a little experiment today of entering all my meals in both apps, and despite the measurements being the same, there's quite a discrepancy in the caloric value between the apps. The other app says 1729 and MFP says 1895. That's quite a big difference! I don't know which one is more accurate but as someone trying to eat in a small deficit and gaining 3lbs instead I'm leaning towards MFP being more accurate! Has anyone else found the same issue between different apps?

cmriverside

On myfitnesspal the food database is crowd-sourced. Every food you use should be checked against a label or a reputable database like USDA (or comparable in other countries.)

The database entries here are often wrong, sometimes just the macros are wrong, but they all need to be verified BY YOU (not by a green check mark) before you use them.

How is the other site's database sourced?

ddsb1111

I think they’re both technically wrong because you’re using someone else’s logging. You can prevent this by weighing and measuring your food on a food scale. I know, it doesn’t sound exciting, but it does take the guess work out. Then it won’t matter what app you’re using.

Btw, I too need to take my own advice, I’m also heading in the wrong direction by guesstimating.

laurachambers86

ddsb1111 wrote: »

I think they’re both technically wrong because you’re using someone else’s logging. You can prevent this by weighing and measuring your food on a food scale. I know, it doesn’t sound exciting, but it does take the guess work out. Then it won’t matter what app you’re using.

Btw, I too need to take my own advice, I’m also heading in the wrong direction by guesstimating.

So a lot of my MFP foods are ones I've previously measured out and saved, although not everything I admit. I might carry on comparing the two for a while and get stricter about measuring things again just for a short period.

This other app also characterises protein bars as "bad" which I also dislike massively but that's another issue!

ddsb1111

Oh yeah, totally. Protein bars are cursed, didn’t you know? As well as anything that’s delicious or brings you joy. Those things are especially bad

cmriverside

What does weighing things have to do with the two databases being different for the same weight item?

In myfitnesspal, many items are just flat entered wrong. They were entered into the shared database by a member on this site. Lots of errors and IMO lots of wishful thinking.

Also, many packaged things change their sizes and recipes over time. The same bread I used to buy is now labeled completely differently.

So, you do need to check against labels periodically - even if you entered the food into the database yourself.

ddsb1111

I agree with what your saying, it’s essentially the same thing.

If she doesn’t weigh and measure her own food neither of those apps would matter because both would be wrong. She can’t get correct “anything” without checking her own nutrition labels, which could be very different than other peoples manual logs. Sorry OP if I wasn’t clear.

AnnPT77

Here's a radical thought: It doesn't matter. Neither is "accurate" in some inviolably absolute sense.

It's all estimates. Yeah, the MFP crowd-sourced database is a factor. But (in the US at least) food product labels are allowed to be off by up to 20% IMU. One apple is sweeter than the next, even at the same weight, so has more calories. Home food scales are probably only accurate to a couple of grams, which is around 18 calories swing, if we're talking about oil/fat.

If you use a consistent data source, and use it carefully (vetting MFP database entries before using them the first time, for example), either the higher estimate or the lower estimate can work fine.

Whaaaat?

Yeah.

No matter what, each person really needs to track consistently for 4-6 weeks (whole menstrual cycles for those to whom that applies), then check the resulting weight change against the expected weight change for the calories eaten. If a person adjusts on the basis of experience, that experience takes into account their typical average logging error rate.

The "calorie goal" is also an estimate. It's basically the average calorie need for people who are similar in the few data items that are input to the estimate (age, weight, height, activity level, etc.). This is true even for fitness trackers, it's just that their estimates are more nuanced/detailed.

Either way, fitness tracker or no, most people are going to be close to the average in calorie expenditure. A few people will be noticeably off, either higher needs or lower. A rare few people will be surprisingly far off, still in either direction. That's the nature of statistical estimates.

This particular statistical estimate is based on BMR/RMR (basal metabolic rate/resting metabolic rate - loosely the amount of calories we'd burn in a coma). Everything else in the calorie needs estimate is built on that BMR/RMR in some way (with the exception of a very few methods of estimating exercise calories). BMR/RMR has a relatively small standard deviation, i.e., a tall, narrow bell curve . . . the implication being that most people are pretty close to average numerically. But a few aren't.

A person who's carefully estimating calories using a particular tool - as long as the tool is reasonably well conceived - is going to tend to have about the same range of error of estimating error on average over time. Therefore, if they think they're eating 1729 calories on average daily, and want to lose more slowly by half a pound a week, they add 250 to 1729 to get a new estimated goal. If they think they're 1895 calories, and want to lose more slowly by half a pound a week, they add 250 calories to get a new estimated goal.

If one is "more accurate", the numeric value of the adjustment amount may need to differ. Over a sweep of time, it won't matter if the base estimate was somewhat low or high, as long as logging practice is consistent, and a person adjusts intake based on individual experience (which they really need to do even if food logging could be exactly exact, because not every human is exactly average).

There's no need to obsess about abstract absolute "accuracy", IMO. It's unachievable. Every single thing is an estimate. They just need to be workable estimates, reasonably consistent over time. That can work.

1895 is 166 calories more than 1729, i.e., 9.6% bigger. MFP's calorie estimate for me (at maintenance) is around 1500 calories (plus exercise). Reality over almost 8 years of logging is that my maintenance calories is around 2000. 2000 is 33% bigger than 1500. I still lose/maintain/gain weight as I expect to.

That happens even though I estimate some meals (wild guess version), have the home scale with its limited accuracy, measure some liquids in cups rather than weighing, eat fruits of varied sweetness, etc. It's close enough. Close enough, on average, over time - that's good enough.

laurachambers86

Thanks all for your insight. I have read before that inaccuracies in counting don't matter as long as you're consistently inconsistent if that makes sense! Also I never knew the data on MFP was crowdsourced, that makes sense now. Not sure if the other app is crowdsourced.

This has given me really useful information in answer to my question, I'm going to weigh my food more consistently for a bit and know that inaccuracies will occur but that its not the end of the world.