http://www.talkstats.com/blog.php Discussion forum for statistics enthusiasts. Get free statistics help or just hang out and talk about statistics. en Mon, 15 Mar 2010 00:32:15 GMT vBulletin 60 http://talkstats.com/images/misc/rss.jpg http://www.talkstats.com/blog.php http://www.talkstats.com/blog.php?b=134 Sat, 13 Mar 2010 02:25:34 GMT Please help! I have the following exercise. Any help it will be appreciated. Thanks in advance. James. :) ================================== ... Please help!

I have the following exercise. Any help it will be appreciated. Thanks in advance. James. :)

==================================

Days: when randomly selecting a day of the week , you are centain to select a day containing the letter y? ]]> sta2023 http://www.talkstats.com/blog.php?b=134 http://www.talkstats.com/blog.php?b=133 Mon, 22 Feb 2010 17:55:37 GMT Here is my question. I hope someone can answer. One of my models has a dichotomous dependent variable and one dummy as the main independent... Here is my question. I hope someone can answer.

One of my models has a dichotomous dependent variable and one dummy as
the main independent variable of interest. I added other dummies
(created from two categorical variables with three groups each) only
to control if the effect of that first dummy changes when these groups
change. I think I should be running logistic regression. But given
that I'm not interested in the direct effect of the control variables
on the dependent variable (only on the direct and conditional effect
of my independent dummy on the response), how should I do it? Running
separate regressions of my categorical variable on the response given
different levels of the control variables wouldn't help, because I
want to control all the variables at once. Is there a special Stata
command that would allow me to control in this way? Is the mfx or
prtab postestimation command meant for this?

I would really appreciate your help with this. ]]> jfgb975 http://www.talkstats.com/blog.php?b=133 http://www.talkstats.com/blog.php?b=132 Fri, 19 Feb 2010 21:45:01 GMT Can someone help me understand and solve below question Medium Fungus Colony Diameters WA 4.5 4.1 4.4 4.0 RDA 7.1 6.8 7.2 6.9... Can someone help me understand and solve below question


Medium Fungus Colony Diameters
WA 4.5 4.1 4.4 4.0
RDA 7.1 6.8 7.2 6.9
PDA 7.8 7.9 7.6 7.6
CMA 6.5 6.2 6.0 6.4
TWA 5.1 5.0 5.4 5.2
PCA 6.1 6.2 6.2 6.0
NA 7.0 6.8 6.6 6.8


(1)
Question:

For laboratory studies of an organism, it is important to provide a medium in which the organism flourishes. The data for this exercise shown in Table above are from a completely randomized design with four samples for each of seven media. The response is the diameters of the colonies of fungus.

A. Perform an analysis of variance to determine whether there are different growth rates among media.

B. Is this exercise appropriate for preplanned or post hoc comparisons? Perform the appropriate method and make recommendations.


I will really appreciate if someone can help me with this question.

Thank you.

Anamika ]]> dz4519 http://www.talkstats.com/blog.php?b=132 http://www.talkstats.com/blog.php?b=131 Tue, 16 Feb 2010 12:06:08 GMT Hello everybody, Im using Kruskal-Wallis to compare group scores on individual likert items (5-point agreement scale) and have researched on which... Hello everybody,

Im using Kruskal-Wallis to compare group scores on individual likert items (5-point agreement scale) and have researched on which post-hoc test(s) to use. The one that comes up most frequently for my situation is of course the Mann-Whiteny U test with a Bonferroni correction to prevent inflation of Type I error rate. However, the problem is I have four different groups (ethnic groups) and therefore need to make six comparisons in total. With the Bonferroni correction this results in a critical level of significanceof .0083 (.05/6=.0083), which is quite strict and could prevent me from observing significant results. Some of the literature I am reading suggests narrowing the comparisons by being selective... for example comparing each group to a baseline group. But I don't have a baseline group since Im comparing four ethnic groups, not experimental conditions.

Does anyone have any suggestions on how to solve this or know if there are any better post-hoc procedures i can use? I was also thinking I could look at the frequencies and possibly make a decision for group comparisons, but thought it might be too much of a subjective method, and i could miss out on the actual important differences.

If i could get help on this, it would be great!!!!! Im working on my Masters thesis and am stressing out majorly.

Thanks in advance!

-Sandra

Some useful info:

-I don't have equal sample sizes for the four different groups ]]> SDuifhuizen http://www.talkstats.com/blog.php?b=131 http://www.talkstats.com/blog.php?b=130 Mon, 08 Feb 2010 23:16:34 GMT Hi,

I've got a dataset of 371 and each entry has 40 variables that dictate which one of six categories it should be put into.

I am trying to find a typical entry for each category but don't really know where to start, having a very limited stats knowledge.

Does anyone have any suggestions as to how I should start, compare variables and see how they affect each other and the category?

Thanks,

Nick ]]> Nick-Nack http://www.talkstats.com/blog.php?b=130 http://www.talkstats.com/blog.php?b=129 Sun, 31 Jan 2010 12:10:59 GMT Hi I need some help with survival analysis. I am doind a retrospective study looking at mortality (event, fatal) and ICD discharges (another... Hi I need some help with survival analysis.

I am doind a retrospective study looking at mortality (event, fatal) and ICD discharges (another event, non fatal).
the problem is that some patients had the ICD discharges before we saw them in clinic (starting of the follow-up). What shall I do? Ignore those ICD discharges?

thanks

Gianni ]]> stata2010 http://www.talkstats.com/blog.php?b=129 http://www.talkstats.com/blog.php?b=128 Sun, 31 Jan 2010 12:07:41 GMT Hi do you know how to calculate and compare mortality by SPSS? thanks Hi do you know how to calculate and compare mortality by SPSS?
thanks ]]> stata2010 http://www.talkstats.com/blog.php?b=128 http://www.talkstats.com/blog.php?b=127 Sun, 31 Jan 2010 12:04:50 GMT Hi Is linear by linear association TEST equivalent to Chi-Square test for trend? thanks Hi

Is linear by linear association TEST equivalent to Chi-Square test for trend?

thanks ]]> stata2010 http://www.talkstats.com/blog.php?b=127 http://www.talkstats.com/blog.php?b=126 Thu, 28 Jan 2010 18:15:18 GMT Hi all, I have been disperately trying to find a way to analyze my data and landed here. I hope someone is there to address my problem. Well, I... Hi all,

I have been disperately trying to find a way to analyze my data and landed here. I hope someone is there to address my problem.

Well, I have two treatments, T1 and T2 randomly applied in six growth chambers (3 each). In each chamber as split plot I have grown 52 genotypes. However, these genotypes belong to 5 different species. Number of genotypes within species are not same. I have collected data from genotypes. I could run SAS Proc mixed to find out the effect of treatments on species and their interaction. Similarly, I could run SAS proc mixed to find out the effect of treatments on genotypes and their interaction. But I would like to know the effect of treatments on species level and on genotypes within each species. It is like partitioning variances in genotypes into 5 species and within species into genotypes. Is there a way to do this with sas procedure? Please help me. ]]> Gautam Pradhan http://www.talkstats.com/blog.php?b=126 http://www.talkstats.com/blog.php?b=125 Tue, 19 Jan 2010 07:51:21 GMT At the moment I am looking at non-verbal representations of language such as emoticons in computer mediated discourse and have compiled a fairly... At the moment I am looking at non-verbal representations of language such as emoticons in computer mediated discourse and have compiled a fairly large Japanese and English corpus.
I am counting non-verbal or paralinguistic cues within these corpora and therefore the corpora may need to be of the same size otherwise my data and findings could be deemed void. For example, if the Japanese corpus if much bigger than the English one then the chances are the more likely that these non-verbal representations will appear. I have tried making the number of sentences the same within each corpora (very time consuming, also defining what a sentence is in online communication can be difficult) and I am also trying to find similar studies that have compared English and Japanese corpora (no luck yet) and to see if here are any reliable representations that state for example that 400 kanji is equal to 1000 English words etc.
What kind of statistical analysis would you recommend which would give me reliable figures regardless of the sizes of my corpora? My study thus far is looking at weblog diaries. The weblog diaries are divided into gender, for example 20 blogs written by women and 20 by men, in both languages (Japanese and English) giving me a total of 80 blogs. From each weblog diary 5 entries are taken giving me a total of 80x5 = 400 entries (200 in each language).
Any ideas or advice in terms of analysis would be fantastic.
Baz.
:D ]]> Baz http://www.talkstats.com/blog.php?b=125 http://www.talkstats.com/blog.php?b=124 Sun, 10 Jan 2010 00:45:47 GMT Hi there,

I am working on my stats for my masters degree and am running a regression and recieve the following message:

"For models with your dependent variable the following variables are constants or have missing correlations... and they were deleted from the analysis"

I'm not sure how to figure out how to fix these variables so they aren't deleted. The rest of the regression goes through just fine. And the two variables it is deleting have the exact same type and amount of information in them. So I'm not sure why they are being deleted.

Please help! Not sure what other info I should pass on...

Thanks ]]> LearningStats6 http://www.talkstats.com/blog.php?b=124 http://www.talkstats.com/blog.php?b=123 Tue, 05 Jan 2010 06:28:15 GMT what analysis should i use? I am doing a research on gender differences in mental health status of asian indian immirants in sydney: impact of age... what analysis should i use?
I am doing a research on gender differences in mental health status of asian indian immirants in sydney: impact of age at migration.
I am using 2 measures GHQ-28 (which has 4 factors) and SCL-90 R (which has 9 factors).
What statistical analysis should i use? any suggestions would be appreciated!! ]]> jignashakhatri http://www.talkstats.com/blog.php?b=123 http://www.talkstats.com/blog.php?b=122 Wed, 30 Dec 2009 16:40:24 GMT Hi,

I was looking into one practical problem of fitting logistic regression to a sales data. I have to assess the stage "Win or Loss" for an organization using some variables like, the account name of the deal (Around 54 different account names are available) and some other variables of this sort. It is very difficult to use dummy or effects coding for this variable as the number of levels are much more in number. can somebody suggest how I can use this nominal variable in fitting Logistic regression. I noticed that this particular variable plays a vital role in winning of a deal.
Thanks,
Yedida ]]> g.yedida http://www.talkstats.com/blog.php?b=122 http://www.talkstats.com/blog.php?b=121 Mon, 28 Dec 2009 18:26:37 GMT :eek: I have a survey with oridinal level data, very unequal sample sizes-330 and 44 , and I cannot assume a normal distribution. Can I use a... :eek: I have a survey with oridinal level data, very unequal sample sizes-330 and 44 , and I cannot assume a normal distribution. Can I use a chi-square? What about Mann-Whitney U ( is the unequal sample size a problem there)? Any other suggestions? ]]> lhart http://www.talkstats.com/blog.php?b=121 http://www.talkstats.com/blog.php?b=120 Fri, 11 Dec 2009 16:40:36 GMT I am using HLM to analyze interaction between individual level and group level outcomes. I have unequal sample size within groups. Using some of the... I am using HLM to analyze interaction between individual level and group level outcomes. I have unequal sample size within groups. Using some of the traditional 'equalizing' methods will reduce my smaple size substantially and I do not want that:)
So, I plan to control for group size. In other words, I want to include group size as covariate. I am just not sure if it should be included in a first level equation, in one of the second level equations, or in all second level equations. I would greatly appriciate any help! ]]> lucia http://www.talkstats.com/blog.php?b=120