[jjc]

quote from the company press release...
"when iPhone owners were asked to perform a text correction task during their sessions, 21% of iPhone owners were not aware of the magnifying glass correction feature although they had owned their iPhone for one month. Participants who did know about the feature clearly loved it, and participants who were new to it indicated that it would be useful in the future"
It's unclear whether they were allowed to use error correction during the main test. Clearly, it is unlikely that the users who moved from other phones to the iPhone did.
I have no idea what a comment like 20 subjects eliminating statistical significance means. That's absurd. A single study with small sample size has poor generalizability and the meaning of non-effects (like a non difference between experience and inexperienced iPhone users) is questionable. But effects that pass statistical tests in such a study generally are more believable, not less. The actual effect size must be larger to come out significant.
As per usual though, this corporate research doesn't release details. How hard would it be to get some folks together to do a little experiment where the full features of the iPhone were used? If auto-correction is turned off that is a serious hindrance to the iPhone because it is all about auto-correction. It would be like comparing manual and automatic transmissions and requiring users to always shift the manual automatic one themselves. Did RIM commission this? /forums/ubbthreads/images/graemlins/smile.gif

[mediafrenzie]

Quote:

---

I don't own an iPhone, but I don't think it takes a rocket scientist to know that humans prefer a tactile response to things they manipulate. The lack of that response is probably part of the blame in the results found in this study. I love my Treo, but when it makes sense to get one, I'll get an iPhone and adjust myself to the virtual keyboard.

---

It's an iPhone thing. The iPhone's ability to second guess you and at times drop in what it BELIEVES you are trying to type even if you've typed out the correct spelling. It's a pain when you're trying to wrap out a fast instruction to a vendor or customer or employee AND only have them reply back with "WHAT?"
Hope that Hells.

[Wondercow]

The people in the study chose a sample of 20 to reduce the statistical error
Reducing the sample size actually increases the chance of error. Not to mention that the confidence level would be very low. Could it be a true representation? Sure. Is it likely? No - the sample size is too small.

[nyip11]

I thought people who like texting don't care about mistakes.

[kboone34]

Quote:

---

The auto correct software is amazing. I would make one change to it though where it would learn the new words you use and that you can also add to the system dictionary. I find that when I type in a word that is unfamiliar to the system but its a word none the less, it tries to correct it with a totally different word. These added features would make it top notch.

---

Trust me, the features you mentioned are on the way. Just hang in there.

[lwdesign]

All this hubbub over a new paradigm buttonless entry system!! Yes, the study was statistically meaningless due to a very small number of people in the test. Also, the guidelines on how the people were selected are missing. Did these folks actually read the manual and learn the iPhone's features? Was auto-correct on or off?
When I see obviously useless studies like this I'm prompted to question the reason for the article and who is paying for it. It is frequent that manufacturers and ad agencies hire writers to ad a PR spin to support specific products or to sow discontent into the ether about competing products. Beware of FUD.
I'm holding out for version 2.0 when the onboard memory moves up to 16GB and higher, then I'll be iPhoning everyone. I'm already with Cingular/AT&T so I don't even have to change providers.

[Hurley42]

Quote:

---

I have no idea what a comment like 20 subjects eliminating statistical significance means. That's absurd. A single study with small sample size has poor generalizability and the meaning of non-effects (like a non difference between experience and inexperienced iPhone users) is questionable. But effects that pass statistical tests in such a study generally are more believable, not less. The actual effect size must be larger to come out significant.

---

"... the significance depends mostly on the sample size ... in very large samples, even very small relations between variables will be significant, whereas in very small samples even very large relations cannot be considered reliable (significant)" (Elementary Concepts in Statistics, StatSoft, Inc. 1984-2003).
Significance is not a term used lightly in statistics. The sample size is too small to infer any real significance considering the size of the population of actual users. This is what is wrong with all those "news" reports that claim x number of people do or prefer Y. Survey results should always be taken with a boulder of salt.
"Why significance of a relation between variables depends on the size of the sample. If there are very few observations, then there are also respectively few possible combinations of the values of the variables, and thus the probability of obtaining by chance a combination of those values indicative of a strong relation is relatively high. Consider the following illustration. If we are interested in two variables (Gender: male/female and WCC: high/low) and there are only four subjects in our sample (two males and two females), then the probability that we will find, purely by chance, a 100% relation between the two variables can be as high as one-eighth. Specifically, there is a one-in-eight chance that both males will have a high WCC and both females a low WCC, or vice versa. Now consider the probability of obtaining such a perfect match by chance if our sample consisted of 100 subjects; the probability of obtaining such an outcome by chance would be practically zero. Let's look at a more general example. Imagine a theoretical population in which the average value of WCC in males and females is exactly the same. Needless to say, if we start replicating a simple experiment by drawing pairs of samples (of males and females) of a particular size from this population and calculating the difference between the average WCC in each pair of samples, most of the experiments will yield results close to 0. However, from time to time, a pair of samples will be drawn where the difference between males and females will be quite different from 0. How often will it happen? The smaller the sample size in each experiment, the more likely it is that we will obtain such erroneous results, which in this case would be results indicative of the existence of a relation between gender and WCC obtained from a population in which such a relation does not exist" (Elementary Concepts in Statistics, StatSoft, Inc. 1984-2003).

[PRoth]

BAH! The test subjects haven't developed their iCalluses yet...

[Moof_in_Charge]

Quote:

---

upon further use I've found I'm much faster at typing complete sentences than ever before. This thing is phenomenal.

---

I cannot agree with you more. I have to say, I no longer type "u" to reference "you", I no longer type "idk", I type "I don't know".
For what it's worth, this so called "study" is a sham, it is a disgrace that anyone in the publishing world would choose to comment or link to a heap of trash which doesn't even adhere to the most basic principles of statistics.
Forget all that mambo jumbo for a second, but "texting" from a phone? What kind of a phone? the type you have to press "2" twice to get "B" or the type that comes with a full keyboard?
How do you derive at 20 users in each group to have similar abilities in spelling anyway? Why didn't they say how long each group has used their particular device to get to where they are now?
Sorry I wasted my time reading this.

[palane]

Quote:

---

I have no idea what a comment like 20 subjects eliminating statistical significance means. That's absurd. A single study with small sample size has poor generalizability and the meaning of non-effects (like a non difference between experience and inexperienced iPhone users) is questionable. But effects that pass statistical tests in such a study generally are more believable, not less. The actual effect size must be larger to come out significant.

---

Well, to do it right, they should have run a T-test. [Statistics geek alert.] It's a way of compensating for small sample size. Average and standard deviation are designed for "infinite" populations. If you deal with a smaller test sample, the margin of error is magnified.
What I love is all of this originally came out of brewing Guinness. So, work with small sample sizes is possible, but the size of the grain of sand you take it with varies. As does your mileage.
BB

[Euge]

I think the way it works is . . . type your non-dictionary word correctly. When the auto-correct word shows up, tap on it to dismiss it. (At first this is not intuitive, but its faster than tapping on it to select it, and its also why the little "x" is beside the word.)
Do this twice and the word will be added to the dictionary.
Dictionary control is rather limited. The only other control you have over it is to clear its contents by going to Settings > General > Reset > Reset Keyboard Dictionary

Quote:

---

Quote:

---

Quote:

---

I would make one change to it though where it would learn the new words you use and that you can also add to the system dictionary.

---

I think it does that ...

---

Really? Doesn't seem like it to me. There isn't an option to add words to dictionary. Does it automatically do it?

---

[Dan Frakes]

Quote:

---

To clarify, I meant they used 20 as opposed to the "samples of 1" that we are getting here. I agree that a sample of 100 would be better from a statistical standpoint, but my point is that it doesn't need to be thousands to be valid.

---

No, but in this case it does need to be more than 20, which isn't all that much better than 1, statistically speaking /forums/ubbthreads/images/graemlins/wink.gif

[Hurley42]

Quote:

---

Quote:

---

I have no idea what a comment like 20 subjects eliminating statistical significance means. That's absurd. A single study with small sample size has poor generalizability and the meaning of non-effects (like a non difference between experience and inexperienced iPhone users) is questionable. But effects that pass statistical tests in such a study generally are more believable, not less. The actual effect size must be larger to come out significant.

---

Well, to do it right, they should have run a T-test. [Statistics geek alert.] It's a way of compensating for small sample size. Average and standard deviation are designed for "infinite" populations. If you deal with a smaller test sample, the margin of error is magnified.
What I love is all of this originally came out of brewing Guinness. So, work with small sample sizes is possible, but the size of the grain of sand you take it with varies. As does your mileage.
BB

---

t tests rule! 1 or 2 tailed though? /forums/ubbthreads/images/graemlins/smile.gif

[Swordmaker]

For this study to really mean anything, the users of the other phones should also have had just one month of experience with their phones.
I also question ANY statistical study that claims that 21% of the 20 iPhone users were unaware of the Magnifying Glass tool... apparently 4.2 of the 20 iPhone users were ignorant? Was one of those users only 20% ignorant and 80% aware of the tool?! That kind of puts their statistical expertise in question.
Another fact that should be mentioned is that Apple is not listed among User Centric's last two quarters clients. However, LG, Microsoft, and Motorola are listed. Why would a for-profit company voluntarilly test a product not manufactured by one of their clients? Did they expend the research dollars out of their desire to help Apple to improve the iPhone? Or, perhaps, we should be looking at another client or two who might have a more nefarious motive? Can we all spell FUD?
This reply was entered entirely with my iPhone.