- Thread starter brownbabygirl
- Start date

1. Put in 95%, as that is the norm.

2. Yes, but you'll likely need some time to get acquainted with it. If it's very simple stuff, Excel will suffice. Otherwise, you may have to look at PSPP (an freeware version of SPSS),**or if you dare, R**.

2. Yes, but you'll likely need some time to get acquainted with it. If it's very simple stuff, Excel will suffice. Otherwise, you may have to look at PSPP (an freeware version of SPSS),

oh and Excel will not suffice as I need t-tests etc

You do the right thing in that you want to get in touch with and talk to a statistician. You also do it right in that you show a willingness to pay for the service the statistician can do for you. (A messy study can be changed to a clear cut study, if planned correct from the start.) If you, your supervisor and your department start a long run cooperation with a statistician, that can turn out to be one of the best investments you have ever made.

Although £50 might be a lot for a student, it is not very much as a consultancy fee. How much would you have to pay for a craftsman for a few hours of work in your country? At TalkStats we usually don't take money.

Sample size?

One way is to use "mean +/- margin of error" like in:

mean +/- 2*s/sqrt(n)

where s is the standard deviation and n is the sample size. Guess a value of s. Increase n until the margin of error is acceptable small.

Good luck!

@brownbabygirl,

You do the right thing in that you want to get in touch with and talk to a statistician. You also do it right in that you show a willingness to pay for the service the statistician can do for you. (A messy study can be changed to a clear cut study, if planned correct from the start.) If you, your supervisor and your department start a long run cooperation with a statistician, that can turn out to be one of the best investments you have ever made.

Although £50 might be a lot for a student, it is not very much as a consultancy fee. How much would you have to pay for a craftsman for a few hours of work in your country? At TalkStats we usually don't take money.

Sample size?

One way is to use "mean +/- margin of error" like in:

mean +/- 2*s/sqrt(n)

where s is the standard deviation and n is the sample size. Guess a value of s. Increase n until the margin of error is acceptable small.

Good luck!

You do the right thing in that you want to get in touch with and talk to a statistician. You also do it right in that you show a willingness to pay for the service the statistician can do for you. (A messy study can be changed to a clear cut study, if planned correct from the start.) If you, your supervisor and your department start a long run cooperation with a statistician, that can turn out to be one of the best investments you have ever made.

Although £50 might be a lot for a student, it is not very much as a consultancy fee. How much would you have to pay for a craftsman for a few hours of work in your country? At TalkStats we usually don't take money.

Sample size?

One way is to use "mean +/- margin of error" like in:

mean +/- 2*s/sqrt(n)

where s is the standard deviation and n is the sample size. Guess a value of s. Increase n until the margin of error is acceptable small.

Good luck!

Hi GG! thanks for this! actually i am now in touch with a consultancy that specifically deal with this sort of thing (biostats help with dissertation). previously did not know they existed.

Sample size?

One way is to use "mean +/- margin of error" like in:

mean +/- 2*s/sqrt(n)

where s is the standard deviation and n is the sample size. Guess a value of s. Increase n until the margin of error is acceptable small.

Good luck!

One way is to use "mean +/- margin of error" like in:

mean +/- 2*s/sqrt(n)

where s is the standard deviation and n is the sample size. Guess a value of s. Increase n until the margin of error is acceptable small.

Good luck!

well mean of what? and sqrt? where do these values come from? i have not collected data yet.

You have not told us anything about what you want to investigate so it is difficult to suggest anything.

Suppose that you have a scale with 5 levels, like: 1, 2, 3, 4, 5 and that the standard deviation is 1.

You need to make the sample size n large enough.

mean +/- 2*s/sqrt(n)

"sqrt" is supposed to be square root.

Suppose the mean of your study would be 3.5 then what is the margin of error? Suppose you used n= 25. then the margin of error would be:

3.5 +/- 2*1/sqrt(25) = 3.5 +/- 0.4 so that the interval would be 3.1 up to 3.9. Would that be precise enough?

Or suppose you had n=100 then it would be 3.5 +/- 2*1/sqrt(100) =3.5 +/- 0.2 with interval 3.3 up to 3.7. Good enough?

It is up to you to choose a big enough sample that is precise enough for your needs.

No, but you need to make the investigation large enough so that it will be precise enough.

You have not told us anything about what you want to investigate so it is difficult to suggest anything.

Suppose that you have a scale with 5 levels, like: 1, 2, 3, 4, 5 and that the standard deviation is 1.

You need to make the sample size n large enough.

mean +/- 2*s/sqrt(n)

"sqrt" is supposed to be square root.

Suppose the mean of your study would be 3.5 then what is the margin of error? Suppose you used n= 25. then the margin of error would be:

3.5 +/- 2*1/sqrt(25) = 3.5 +/- 0.4 so that the interval would be 3.1 up to 3.9. Would that be precise enough?

Or suppose you had n=100 then it would be 3.5 +/- 2*1/sqrt(100) =3.5 +/- 0.2 with interval 3.3 up to 3.7. Good enough?

It is up to you to choose a big enough sample that is precise enough for your needs.

You have not told us anything about what you want to investigate so it is difficult to suggest anything.

Suppose that you have a scale with 5 levels, like: 1, 2, 3, 4, 5 and that the standard deviation is 1.

You need to make the sample size n large enough.

mean +/- 2*s/sqrt(n)

"sqrt" is supposed to be square root.

Suppose the mean of your study would be 3.5 then what is the margin of error? Suppose you used n= 25. then the margin of error would be:

3.5 +/- 2*1/sqrt(25) = 3.5 +/- 0.4 so that the interval would be 3.1 up to 3.9. Would that be precise enough?

Or suppose you had n=100 then it would be 3.5 +/- 2*1/sqrt(100) =3.5 +/- 0.2 with interval 3.3 up to 3.7. Good enough?

It is up to you to choose a big enough sample that is precise enough for your needs.

fyi My study is looking at how aware pregnant women are of their sickle cell status and the factors that contribute to these.

thanks!