Help needed - choosing the proper statistical test

[ssam]

What Statistical tests can I use for my data? I've represented my data as a bar graph with stdev error bars. I need a p value (i.e. p = 0.05)? Should I use T-test, chi sq, anova? I know skme of these won't be applicable but could someone what method I could use to analyse this data? This is my data...

	Resting HR	HR after 1 min of vigorous exercise.	HR after 5 mins of rest
Repeat 1	81	127	84
Repeat 2	86	130	88
Repeat 3	77	121	81

 

[Deleted member 4993]

What Statistical tests can I use for my data? I've represented my data as a bar graph with stdev error bars. I need a p value (i.e. p = 0.05)? Should I use T-test, chi sq, anova? I know skme of these won't be applicable but could someone what method I could use to analyse this data? This is my data...

	Resting HR	HR after 1 min of vigorous exercise.	HR after 5 mins of rest
Repeat 1	81	127	84
Repeat 2	86	130	88
Repeat 3	77	121	81

Please show us what you have tried and exactly where you are stuck.

Please follow the rules of posting in this forum, as enunciated at:

READ BEFORE POSTING​

Please share your work/thoughts about this problem.

 

[BigBeachBanana]

What Statistical tests can I use for my data? I've represented my data as a bar graph with stdev error bars. I need a p value (i.e. p = 0.05)? Should I use T-test, chi sq, anova? I know skme of these won't be applicable but could someone what method I could use to analyse this data? This is my data...

	Resting HR	HR after 1 min of vigorous exercise.	HR after 5 mins of rest
Repeat 1	81	127	84
Repeat 2	86	130	88
Repeat 3	77	121	81

What are you trying to test?

 

[ssam]

I'm looking for the most appropriate statistical techniques to use to analyse the data. I'm not a maths student so I'm unfamiliar with the appropriate techniques to use. I'm looking for recommendations really - Maybe use a student t-test to see if resting heart rate (HR) and HR after 5 mins are significantly different. Perhaps an ANOVA test to compare all 3? But then would I need the T-test? Basically, I'm looking for the most appropriate methods to use and then ill look up how to use them. Thank you.

 

[Deleted member 4993]

I'm looking for the most appropriate statistical techniques to use to analyse the data. I'm not a maths student so I'm unfamiliar with the appropriate techniques to use. I'm looking for recommendations really - Maybe use a student t-test to see if resting heart rate (HR) and HR after 5 mins are significantly different. Perhaps an ANOVA test to compare all 3? But then would I need the T-test? Basically, I'm looking for the most appropriate methods to use and then ill look up how to use them. Thank you.

As I see it, you have 3 independent data points - i.e. your population size is 3.

Is that correct view of your data-set?

 

[ssam]

As I see it, you have 3 independent data points - i.e. your population size is 3.

Is that correct view of your data-set?

Hi Subhotosh,

Yes, I'd assume so. Does this mean n=3? Apologise but I have little experience in statistics and so the reason for posting my query is to gain advice from people more knowledgeable in the field.

I have presented my data in a bar graph (see below).



My values for a t-test can also be seen below...

Rest vs recover	0.2701442​
Rest vs exercise	0.0001692​
Recovery vs exercise	0.0002356​

I was wondering if it would be more appropriate to conduct an ANOVA test to compare all 3 variables at once?

I need help with relating these techniques (or any other more relevant statistical techniques you could suggest) with my p value which ive taken as p<0.05.

Any suggestions would be greatly appreciated,

Many thanks.

 

[BigBeachBanana]

How large is your data set? Is it only those 3 or this is only representative of your data set?
Also when you say repeat1,2,3. Is it on the same test subject or different test subjects?

 

[ssam]

How large is your data set? Is it only those 3 or this is only representative of your data set?
Also when you say repeat1,2,3. Is it on the same test subject or different test subjects?

Hi bigbeachbannan,
So the repeats were conducted on 3 different days (repeat 1 = day 1, repeat 2 = day 2, repeat 3 = day 3) measuring the 3 parameters mentioned (resting HR, HR straight after exercise and HR 5 mins after exercise). The data was retrieved from 1 subject (me). The data presented is raw data measured and not representative of a data set. So I'm working with very little data but this is what I have to work with. Hope that help explain this.

 

[BigBeachBanana]

Hi bigbeachbannan,
So the repeats were conducted on 3 different days (repeat 1 = day 1, repeat 2 = day 2, repeat 3 = day 3) measuring the 3 parameters mentioned (resting HR, HR straight after exercise and HR 5 mins after exercise). The data was retrieved from 1 subject (me). The data presented is raw data measured and not representative of a data set. So I'm working with very little data but this is what I have to work with. Hope that help explain this.

I’m trying to understand the thesis of your study.Like what question are you trying to answer? Are you trying to see if there’s a correlation between HR at rest, after 1 min and 5 min?

 

[ssam]

Realistically I know resting HR and recovery HR are not significantly different but they are both significantly different to exercise HR - so what I'm looking for is to describe this in a figure legend but using the correct stat technique. Would you suggest attest, ANOVA, etc... ?

 

[Deleted member 4993]

As I see it, you have 3 independent data points - i.e. your population size is 3.

Now I understand - actually your population size is 1 - with 3 repeats. What do you plan to do with the result? Are planning to present the data for a project - or science fair or what?

 

[BigBeachBanana]

Realistically I know resting HR and recovery HR are not significantly different but they are both significantly different to exercise HR - so what I'm looking for is to describe this in a figure legend but using the correct stat technique. Would you suggest attest, ANOVA, etc... ?

1. I'm frankly still don't understand the purpose of your study.
2. To conduct a statistical test, there are specific criteria your data needs to meet.
   • Your observations are independent of each other, and yours all come from 1 test subject.
   • The observations are drawn from a normally distributed population, but you only have n=1
3. Comparing resting HR and recovery HR doesn't provide any insight. What question does this answer? A meaningful comparison would be, for example, recovery HR for biking vs. recovery HR for rock climbing.
4. I'm unsure of the expectation from your instructor, but I'd dismiss the credibility of this study within a heartbeat with three observations (no punt intended)
5. The most you can do is descriptive statistics such as mean, median, mode, variance, kurtosis, interquartile, etc... to describe your data. But what you're looking for is inferential statistical, which you don't have credible data and a clear question you're trying to answer; thus, you can't infer anything.
6. Another option would be to reframe your study. How long does it take after, say biking (or whatever activity you did to collect your data), for the HR to recover?

 

[ssam]

1. I'm frankly still don't understand the purpose of your study.
2. To conduct a statistical test, there are specific criteria your data needs to meet.
   • Your observations are independent of each other, and yours all come from 1 test subject.
   • The observations are drawn from a normally distributed population, but you only have n=1
3. Comparing resting HR and recovery HR doesn't provide any insight. What question does this answer? A meaningful comparison would be, for example, recovery HR for biking vs. recovery HR for rock climbing.
4. I'm unsure of the expectation from your instructor, but I'd dismiss the credibility of this study within a heartbeat with three observations (no punt intended)
5. The most you can do is descriptive statistics such as mean, median, mode, variance, kurtosis, interquartile, etc... to describe your data. But what you're looking for is inferential statistical, which you don't have credible data and a clear question you're trying to answer; thus, you can't infer anything.

So this isn't a study it's just part of an assignment - I need to present this data in the best way possible and had a figure legend that accompanies it. Part of what makes up the Figure legend will be me including the information p<0.05. To include p<0.05 in the legend I must have performed some sort of statistical test on this very basic data otherwise if I just put p<0.05 in the figure, it will make no sense. I need to put p<0.05 and then state what statistical test I used to analyse the data. Going back to my original question, I don't know what test is most appropriate to allude to in the figure's legend and I'm asking for suggestions. I hope that makes science now. This data is rubbish and extremely limited, granted, it won't make your heart skip a beat (pun intended) but I need to work with what I've got and somehow include a statistical test that would allow me to include p<0.05 in the legend. The answer may be very easy but again I'm not a maths student so these things go over my head. Thank you

 

[BigBeachBanana]

- I need to present this data in the best way possible and had a figure legend that accompanies it.

This goes back to what I stated before. The assignment aims to summarize/describe the data, i.e., descriptive statistics. Performing statistical test falls under inferential statistics. Is this what you want to do, or the assignment requires you to do it. You can present the data with a triple bar graph that compares the HR between different time intervals and days, with a legend that would satisfy the assignment requirement.
For example:

Think about how you could create one for your data.

 

[ssam]

This goes back to what I stated before. The assignment aims to summarize/describe the data, i.e., descriptive statistics. Performing statistical test falls under inferential statistics. Is this what you want to do, or the assignment requires you to do it. You can present the data with a triple bar graph that compares the HR between different time intervals and days, with a legend that would satisfy the assignment requirement.
For example:
View attachment 31646
Think about how you could create one for your data.

I've tried representing the data multiple ways...

Part of the assignment is to present in the appropriate format and include all statistics, error bars and P values where appropriate. To include P values ill have to do some sort of stats (maybe very simple) to get a p value - First which format do you think is best? With 2 and 3 i can incle error bars as each point is an average of 3 results, but for 1 (the way you suggested) can cant include error bars. Ignore the axis for fig 2. Still looking for a reason to incle p<0.005 in the legend?

 

[BigBeachBanana]

I've tried representing the data multiple ways...
View attachment 31647View attachment 31648View attachment 31649
Part of the assignment is to present in the appropriate format and include all statistics, error bars and P values where appropriate. To include P values ill have to do some sort of stats (maybe very simple) to get a p value - First which format do you think is best? With 2 and 3 i can incle error bars as each point is an average of 3 results, but for 1 (the way you suggested) can cant include error bars. Ignore the axis for fig 2. Still looking for a reason to incle p<0.005 in the legend?

You can add error bars to the first graph. See this.(Youtube link)

 

[ssam]

You can add error bars to the first graph. See this.(Youtube link)

How is that possible? they are individual values - there cant be stdev error bars on individual values?

 

[BigBeachBanana]

How is that possible? they are individual values - there cant be stdev error bars on individual values?

My apologies. I've reread what I wrote and knew why you've might misunderstand what I meant. What I meant was you can add a 4th category to the first graph. Essentially combine 1st and 3rd graph into one, so repeat 1, 2, 3, and averages. On the averages, you can add the error bars.

As far as the p-value, you can conduct a two-tailed comparison of two population means. You can visually see that from your 3rd graph. If the error bars overlap then there's no significant difference (Resting vs. Recovery), whereas when the error bars do not overlap then there's a significant difference (Resting vs Excercise and Excercise vs.Recovery)

 

[ssam]

My apologies. I've reread what I wrote and knew why you've might misunderstand what I meant. What I meant was you can add a 4th category to the first graph. Essentially combine 1st and 3rd graph into one, so repeat 1, 2, 3, and averages. On the averages, you can add the error bars.

As far as the p-value, you can conduct a two-tailed comparison of two population means. You can visually see that from your 3rd graph. If the error bars overlap then there's no significant difference (Resting vs. Recovery), whereas when the error bars do not overlap then there's a significant difference (Resting vs Excercise and Excercise vs.Recovery)

Ok so I've decided to use a t-test to compare results.
Ive you look at my main graph, what i want to get through is that resting & recovery are significantly different to exercise but resting and recovery are not significantly different to each other..

Ive use the t-test function on excel but I'm not entirely sure if I've input the correct data...



This is what it ask you to enter - i saw online somewhere to assume talis and type = 2 but im unsure if this is correct - could you provide any clarification?

In the figure legend ill say ive taken p<0.05 and these are my p values. I just need to check the p values ive calculated are correct. ive looked at some online calculators which produce different p values to what i have obtained. Could you help please ?

 

[BigBeachBanana]

Ok so I've decided to use a t-test to compare results.
Ive you look at my main graph, what i want to get through is that resting & recovery are significantly different to exercise but resting and recovery are not significantly different to each other..

Ive use the t-test function on excel but I'm not entirely sure if I've input the correct data...
View attachment 31657
View attachment 31660

This is what it ask you to enter - i saw online somewhere to assume talis and type = 2 but im unsure if this is correct - could you provide any clarification?

In the figure legend ill say ive taken p<0.05 and these are my p values. I just need to check the p values ive calculated are correct. ive looked at some online calculators which produce different p values to what i have obtained. Could you help please ?

What you computed is the t-statistic, not the p-value. You need to look up the corresponding p-value from the t-distribution.