Peter Ellis

9,566 reputation

1043

bio	website	med.govt.nz/…
	location	Wellington, New Zealand
	age	40
visits	member for	1 year, 5 months
	seen	4 hours ago
stats	profile views	658

Manager of Sector Performance at the New Zealand Ministry of Business, Innovation and Employment. Master of Applied Statistics from Australian National University. Fourteen years of experience in overseas aid, mainly in program evaluation and performance information but also program management, including as Counsellor Development Cooperation in the Australian Embassy in Dili, East Timor.

My team's work includes complex surveys and other data collection and analysis to help evidence-informed policy concerning the tourism industry, science and innovation system, and the sectoral and regional drivers of economic growth in New Zealand.

	9,566 reputation	bio	website	med.govt.nz/…	visits	member for	1 year, 5 months
1043 badges		location	Wellington, New Zealand		seen	4 hours ago

summary answers questions tags badges favorites bounties reputation activity

1,231 Actions

suggestions reviews revisions comments badges posts accepts all

Apr 13	answered	Max-Diff Using Rank-Order Data?
Apr 13	comment	Is LOESS the appropriate way of visualizing my RT data? ah, are these repeat measurements on subjects that get multiple trials each?
Apr 13	revised	Is LOESS the appropriate way of visualizing my RT data? added ggplot2 tag
Apr 13	comment	Is LOESS the appropriate way of visualizing my RT data? Could you clarify how time comes into the experiment - you say you expected differences "to become larger as blocks advanced" - what does this mean? Also, are you plotting the points using geom_point() or geom_jitter() as well as stat_smooth() - if so it should help for a reality check. If there's too much data to see try something like + geom_point(alpha=0.1) to add transparency.
Apr 13	comment	Hypothesis test on data with confounding spatial clustering Is there a danger that an underlying spatial feature is a significant factor - beyond just the issue that close cells have similar scores? As the treatment cells are clustered in space, if they are unluckily in a zone that happens to impact on the response, it won't be possible to distinguish the effects.
Apr 13	comment	Does R predict or forecast on an arimax model? @b2amen if you found the answer useful can you select the tick next to it to indicate your question is resolved? thanks.
Apr 13	revised	Does R predict or forecast on an arimax model? formatting to make code fit on screen
Apr 13	comment	Does R predict or forecast on an arimax model? I've put the predict() code into the question - more likely to be seen there than in the comments. BTW you can also use the clickable icons to format a selected part of your question or answer as code.
Apr 13	revised	Does R predict or forecast on an arimax model? inserted predict() code into question (was just a comment)
Apr 13	comment	weighted regression with dummy variables The fact that you only have dummy variables for predictors does not make any difference. So this question reduces to references on weighted regression which would turn this into "big list" question - there are so many, where to start? And as @Adamo asks, we need to know if the weighting is because you are weighting to population, frequency weighting because you have multiple observations of some points, or weighting for some other reason eg to account for variable variance.
Apr 13	revised	How to calculate the standard error of the marginal effects in interactions (robust regression)? formatted equation
Apr 13	answered	PCR or PCSR (Principal Component Score Regression)?
Apr 13	answered	What statistical test do I need to see if there is a correlation between two different instruments?
Apr 12	comment	What statistical test do I need to see if there is a correlation between two different instruments? Is there a standard way of processing the results from each instrument into one or two dimensions eg a "self confidence index", "satisfaction index" and "intuitiveness index" - or are you going to look for connections at the level of individual questions? Also, I assume you will have a sample of subjects who completed both instruments?
Apr 10	comment	Determining if awards are fair I suspect the question was downvoted because, although written very clearly, it is a straightforward example of an absolutely classic situation that would feature in most introductory stats texts, but you hadn't indicated what you had tried before asking the question. People may have suspected it was untagged 'homework' or self-learning. I presume my answer got a downvote because it is too brief - doesn't explain why my answer is appropriate or how it works(and I agree, I just don't have time to expand on it properly just now, and the expansion can be found in many texts anyway).
Apr 9	comment	convert annual data to quarterly frequency My point is that it isn't possible to disaggregate annual to quarterly. That is, there is no possible way to recreate the first of my two plots if all you have is the data for the second. Anyway, good luck with your analysis.
Apr 9	answered	Determining if awards are fair
Apr 7	comment	Why can't I trim the the dependent variable in a regression? Or can I? thanks for the clarification about terminology @richardh - when I read your question first I assumed you meant for some reason it was common to discard large outliers (right hand side of the distribution) but not small ones (left hand side of the distribution).
Apr 7	answered	Mixed model in simple english
Apr 7	comment	weighted multiple linear regression @JeffG. you'll get a better answer if you explain exactly what you're doing, including at a minimum what the "other regression technique" is.